Comment on LLM function calls don't scale; code orchestration is simpler, more effectiveparentComments−nitwit0051yI've seen the suggestion it's because it's been trained on a lot of HTML, but the GPT docs suggest using markdown as a default choice, which is relatively less common.
Comments
I've seen the suggestion it's because it's been trained on a lot of HTML, but the GPT docs suggest using markdown as a default choice, which is relatively less common.